Introduction

Although the earliest models of receptive fields of simple cells in V1 were based primarily on excitation from afferent LGN fibers (Hubel and Wiesel, 1962), many researchers have since recognized the importance of inhibition from inside (DeAngelis et al., 1992; Morrone et al., 1982; Bonds, 1989) and outside (DeAngelis et al., 1994; Nelson and Frost, 1978; Blakemore and Tobin, 1972; Knierim and Van Essen, 1992; Hubel and Wiesel, 1968) the classical receptive field (CRF) in shaping the response properties of V1 neurons. When a neuron is presented with and responds to an optimally oriented stimulus, it is possible to reduce its response by superimposing an orthogonal mask stimulus or by placing a parallel stimulus outside the CRF. These two phenomena are commonly known as cross-orientation suppression and iso-orientation surround suppression, respectively.

The selectivity of cross-orientation suppression has been tested in several studies. Suppression appears to originate from within the CRF (DeAngelis et al., 1992), and is strongest when the size of the mask is equal to or smaller than that of the excitatory stimulus. It is broadly selective for spatial frequency (Bauman and Bonds, 1991; DeAngelis et al., 1992) and temporal frequency (Freeman et al., 2002; Allison et al., 2001). However, it has little or no selectivity for orientation (DeAngelis et al., 1992; Bonds, 1989).

Three mechanisms have been proposed to account for cross-orientation suppression. (1) Inhibitory phenomena in V1 might act through a common pool of inhibition among striate neurons (Heeger, 1992; Bonds, 1989). (2) An extrastriate source for cross-orientation suppression was proposed by Allison et al. (2001). They studied the temporal frequency selectivity of the mask, and found that the peak frequency (7.0$\pm$2.6 Hz) was higher than that for a preferred grating stimulus in cat area 17 (3.8$\pm$1.5 Hz), but similar to that in area 18 (7.2$\pm$3.1 Hz). They found a similar result for cutoff temporal frequency. (3) Recently, synaptic depression in thalamocortical synapses has been proposed to underly this form of suppression. Freeman et al. (2002) found a high cutoff temporal frequency for the mask grating (similar to Allison et al. (2001)). They were able to induce suppression with gratings drifting at rates that were too high to drive most area 17 neurons in the cat. In addition, they were not able to adapt out the suppression with orthogonal gratings that were sufficiently strong to adapt the excitatory stimulus. This led them to suggest a non-cortical mechanism underlying cross-orientation suppression.

Like cross-orientation suppression, the suppressive signals which originate from outside the CRF have been studied extensively. In most cells, the magnitude of surround suppression is strongest at orientations near the cell's optimal (DeAngelis et al., 1994; Sengpiel et al., 1997). Similarly, the most suppressive stimuli are those which match the preferred spatial frequency (DeAngelis et al., 1994; Morrone et al., 1982). For both orientation and spatial frequency, the surround mechanism shows broader tuning than the excitatory CRF.

The nature of the circuitry which underlies surround suppression has been discussed in many studies. There have been two main proposals for this circuitry. (1) Extensive horizontal axonal projections within V1 have been implicated as a possible substrate (DeAngelis et al., 1994; Nelson and Frost, 1978; Blakemore and Tobin, 1972; Knierim and Van Essen, 1992; Hubel and Wiesel, 1968). (2) Several studies that found a delay in neuronal signals for contextual modulation (Knierim and Van Essen, 1992; Zipser et al., 1996; Lamme, 1995) led to suggestions that surround mechanisms may operate through feedback from higher cortical areas (Levitt and Lund, 2002; Angelucci et al., 2002).

Heeger (1992) formalized a model of inhibition in which a striate cortical cells normalize their responses to stimulus contrast through mutually inhibiting each other. He further proposed that this mutual inhibition might underly several suppressive mechanisms within primary visual cortex, including suppression from the surround and from orthogonal masks within the CRF. If these two forms of suppression work through the same mechanism, it would seem likely that their timing would reflect this similarity. An assessment of the latencies involved in surround suppression and cross-orientation suppression can provide further evidence about whether a common mechanism is involved.

To date, we are unaware of any reports of the time course of cross-orientation suppression, although for surround suppression it has been addressed in some studies (Nothdurft et al., 1999; Knierim and Van Essen, 1992) and recently examined more extensively (Bair et al., 2003). The circuitry which underlies cross-orientation suppression is not known, and in fact has been the subject of much interest (Freeman et al., 2002; Heeger, 1992; Bonds, 1989; Lauritzen et al., 2001). By studying the timing of this suppressive mechanism, we can place important constraints on the possible circuitry, and supply data to models about how it occurs (Freeman et al., 2002; Lauritzen et al., 2001).